Shoulder MR arthrography of the posterior labrocapsular complex in overhead throwers with pathologic internal impingement and internal rotation deficit
- First Online:
- Cite this article as:
- Tuite, M.J., Petersen, B.D., Wise, S.M. et al. Skeletal Radiol (2007) 36: 495. doi:10.1007/s00256-007-0278-6
- 411 Views
To determine if overhead-throwing athletes with internal impingement pain and internal rotation deficit have thickening of the posterior inferior labrocapsular complex on MR arthrogram images.
Materials and methods
This study was approved and a waiver of consent granted by our institutional review board. Twenty-six overhead-throwing athletes with internal impingement pain and internal rotation deficit, and 26 controls who had undergone MR arthrograms, were retrospectively examined. The MR studies were combined and read in a blind fashion. On an axial image through the posteroinferior glenoid rim, the readers measured the labral length, capsule-labrum length, and the posterior recess angle. A t-test was used to determine statistical significance.
The mean labral length was 4.9 mm [standard deviation (SD) 1.4 mm] for the controls, and 6.4 mm (SD 1.6 mm) for the athletes (P = 0.001). The mean capsule-labrum length was 5.4 mm (SD 2.1 mm) for the controls, and 8.8 mm (SD 2.9 mm) for the athletes (P < 0.001). The mean posterior recess angle measured 65° (SD 27°) for the controls and 94° (SD 38°) for the athletes (P = 0.002).
Overhead-throwing athletes with internal impingement pain and internal rotation deficit tend to have a thicker labrum and a shallower capsular recess in the posterior inferior shoulder joint than do non-overhead-throwing athletes. In many, the posteroinferior capsule is also thickened. These MR findings should alert the radiologist to closely inspect the posterior cuff and posterosuperior labrum for the tears associated with internal impingement.
KeywordsShoulderMRIOverhead-throwing athletic injuries
There are multiple causes of shoulder pain in the overhead-throwing athlete, including rotator cuff and biceps tendonosis, glenohumeral instability, and rotator interval injury [1–5]. Another cause of shoulder pain is pathologic internal impingement, where the athlete experiences pain in the late cocking and early acceleration phase of throwing [6–10]. These athletes also suffer a decrease in throwing velocity, which negatively affects their athletic performance. Several authors have described the MR and MR arthrographic findings common in pathologic internal impingement, including tears of the posterosuperior labrum and posterior rotator cuff [1, 4, 11–13].
There are two main biomechanical explanations for why some overhead-throwing athletes develop internal impingement pain and tears. The first explanation is that the labrum, cuff and well-innervated joint capsule are repetitively pinched between the greater tuberosity and the glenoid rim during the throwing motion, which leads to fraying, tears and pain [7, 8, 10]. The second theory is that the overhead-throwing athlete first develops a tight posterior inferior capsule, which changes the glenohumeral contact point in the late cocking position and leads to progressive stretching of the anterior capsule [6, 10]. These changes intensify what otherwise would be gentle pinching of the labrum, cuff and joint capsule between the greater tuberosity and glenoid rim, as well as cause twisting shear stress tears of the posterior cuff and labrum.
The evidence for this second theory is that many athletes with pathologic internal impingement clinically have a significant glenohumeral internal rotation deficit (GIRD) compared with the contralateral shoulder, presumably due to a tight posterior inferior capsule [6, 10, 14]. At arthroscopy, the posterior inferior capsule is often thickened in symptomatic elite professional baseball pitchers who have been pitching for many years and who do not improve with selective posteroinferior capsule stretching exercises [6, 15]. The posteroinferior capsular recess is also often small and contracted, due to the tight posteroinferior capsule .
We wondered if milder changes of the posteroinferior capsule and adjacent labrum were present in younger baseball pitchers and in other overhead-throwing athletes who had internal impingement pain and GIRD. Recognizing this finding on imaging might be useful in predicting who is at risk for developing a posterior-type superior labrum anterior to posterior (SLAP) or cuff tear or at risk for requiring surgical posteroinferior capsule release . The purpose of this study was to determine if thickening of the posterior inferior labrocapsular complex was present on MR arthrographic images in overhead-throwing athletes with internal impingement pain and GIRD who were not elite professional baseball pitchers.
Materials and methods
This study was approved and a waiver of consent granted by our institutional review board.
Twenty-six consecutive patients between 2000 and 2005 who were overhead-throwing athletes with both internal impingement pain and GIRD, and who underwent a shoulder MR arthrogram, were included in the study. None of the patients was a professional baseball pitcher. There were 21 men and five women, with a mean age of 23 years (range 16–38 years). Eleven of the patients were high school, college, or amateur league baseball pitchers, seven were softball or baseball position players, three played collegiate volleyball, two collegiate tennis, two were javelin throwers, and one was a high school football quarterback.
All patients developed their pain initially while performing an overhead throwing or serving motion and continued to have subsequent pain in the late cocking or early acceleration phase of throwing. The patients were examined clinically by one of three shoulder specialists in our sports medicine clinic (two orthopedists and one sports medicine physician). All patients had a clinical diagnosis of pathologic internal impingement, defined as posterosuperior glenohumeral joint pain with the arm in the abduction external rotation (ABER) apprehension position, which was relieved on the relocation test of Jobe . The other inclusion criterion was decreased glenohumeral internal rotation in the effected dominant shoulder. Internal rotation was measured in 19 of the 26 patients with the patient supine and the humerus abducted to 90° and the arm maximally internally rotated while the scapula was stabilized against the tabletop . The average internal rotation deficit in the dominant shoulder was 26°, compared with that of the contralateral shoulder (range 10–45°). In seven patients internal rotation was determined by the ‘back scratch test’, where the patient adducts and internally rotates the arm behind the back and reaches to the highest vertebral body level . These patients could reach an average of only 4.7 vertebral bodies lower on their effected side (range 2–8 vertebra levels).
In addition to the diagnosis of pathologic internal impingement, the indication listed for the MR arthrogram was internal impingement cuff or labral tear in ten, SLAP tear in nine, SLAP tear plus biceps tendonosis in three, posterior labral tear in two, anterosuperior labral tear in one, and rotator cuff impingement in one. On the original MR interpretation, 15 patients had shown a tear or fraying of the posterior cuff or posterosuperior labrum included in the report.
A control group was included in the study and consisted of 26 similar age and gender patients who were not overhead-throwing athletes, did not have pathologic internal impingement by history or physical examination, and had undergone a shoulder MR arthrogram, that was interpreted as normal, during a portion of the same time period. The control group included 19 men and seven women, with a mean age of 24 years (range 16–39 years). The clinical diagnosis at the time of the MR arthrogram was SLAP tear in nine, SLAP tear plus biceps tendonosis in one, instability in eight, rotator cuff impingement/strain in six, scapulothoracic dysfunction with cuff weakness in one, and subscapularis injury in one.
For the MR arthrogram, informed written consent was obtained, and a 22-gauge needle was placed into the joint under fluoroscopic control; 12 ml to 15 ml of a 1:200 dilution of gadolinium (Omniscan, Amersham Health, Princeton, NJ, USA) was instilled into the joint. The patient was escorted immediately to the MR scanner. All MR arthrogram images were obtained on a 1.5 T scanner (GE Medical Systems, Milwaukee, WI, USA) using a phased array shoulder coil (IGC-Medical Advances, Milwaukee, WI, USA).
Oblique sagittal fat-suppressed T1-weighted images (TR/TE=633/18 ms, one excitation) were obtained with a 4 mm section thickness and 1 mm interslice gap. Oblique axial images were localized on a sagittal image so that they were perpendicular to the long axis of the glenoid fossa. These were fat-suppressed T1-weighted images (667/17 ms, 1.5 excitations) with a 3 mm section thickness and 1 mm interslice gap. Additional sequences that were obtained but not used in this study were oblique coronal fat-suppressed T1-weighted (633/18 ms, one excitation), and fat-suppressed fast spin-echo (FSE) T2-weighted images (2,033/60 ms, echo train length six, and three excitations) with a 4 mm section thickness and 1 mm interslice gap. In addition, ABER oblique axial fat-suppressed T1-weighted images (700/18 ms, 1.5 excitations) with a 4 mm section and 1 mm interslice gap were obtained, using paired 5 in. coils. All images were done with a 14 cm field-of-view and a 256 × 192 pixel matrix.
The MR images were reviewed on a picture archiving communication system (PACS) workstation (McKesson Information Solutions LLC, Richmond, BC, Canada). The internal impingement/GIRD and control patients’ examinations were combined, and the studies were evaluated in alphabetical order by last name with the readers blind to the patient’s medical history. The measurements were made separately by a musculoskeletal fellow and a musculoskeletal radiologist with 14 years experience. The musculoskeletal radiologist re-measured the images 1 month later to determine intra-observer variability.
We determined the labral length by measuring from the outer edge of the posterolateral corner of the glenoid rim subchondral bone to the lateral tip of the labrum. The undercutting hyaline cartilage thickness was included in the measurement. The measurement was made to the nearest tenth of a millimeter using the linear distance tool on the PACS workstation.
‘Thick-capsule labrum length’
If the joint capsule appeared to insert near the tip of the labrum, we measured the ‘thick-capsule labrum length’ by drawing a straight line from the glenoid rim through the labrum and capsule to the point where the capsule thinned out to become 1–2 mm thick, similar to normal capsule more laterally (Fig. 2). If the capsule was normal and thin at the attachment, the ‘thick-capsule labrum length’ was recorded as being equal to the labral length.
Posterior recess angle
The posterior recess was defined as the joint fluid posterior and medial to the posterior labrum. The angle formed by the margins of the posterior recess was measured to the nearest degree using the angle function on the PACS workstation. This was an acute angle in those patients with a well-formed joint recess where the anterior margin of the recess was the posterior labrum, and, in these patients, their ‘thick-capsule labrum length’ also equaled their labral length. In those patients where the posterior capsule appeared to insert near the tip of the labrum, the anterior margin of the shallow recess was typically also the joint capsule and an obtuse angle was measured.
We also measured the degree of external/internal rotation of the humeral head, using the technique of Davis et al. .
The ages of the two groups were compared with a t-test, and the genders were compared with a Fisher’s exact test. The labral length, ‘thick-capsule labrum length’, and posterior recess angle were compared between the internal impingement/GIRD group and the controls, using a two-sample t-test separately for each reader and repeated measures analysis of variance (ANOVA) combining data from reader one and the first measurements by reader two. We also computed the areas under the curve (AUCs), based on receiver operating characteristic (ROC) analysis for each measurement.
Inter-observer and intra-observer variability was assessed using Pearson product moment correlation to assess linear association of the measurements. For between-reader assessment, the analysis was based on the first measurement for reader 2. To assess the absolute difference in measurements, we computed mean percent difference between measurements, defined as the absolute difference of the measurements divided by their average. In all analyses, a P value less than 0.05 was considered significant.
There was no statistical difference between the two groups in the distribution of ages (P = 0.32) or genders (p = 0.28).
Mean labrocapsular measurements at the posterior inferior glenoid rim
Internal impingement with internal rotation deficit athletes
Labral length-mm (SD)
Thick-capsule labrum length-mm (SD)
Posterior recess angle-degrees (SD)
The ‘thick-capsule labrum length’ varied between 2.3 mm and 11.7 mm for the controls and 2.9 mm and 16.5 mm for the internal impingement/GIRD athletes. The mean length was 5.4 mm (SD 2.1 mm) for the controls and 8.8 mm (SD 2.9 mm) for the internal impingement/GIRD athletes (P < 0.001). The ‘thick-capsule labrum length’ was the same as the labral length for reader 1 in 21 of 26 controls (81%) and for reader 2 in 23 of 26 controls (88%). For the internal impingement/GIRD athletes, the ‘thick-capsule labrum length’ and labral length were the same for reader 1 in 11 of 26 athletes (42%) and for reader 2 in nine of 26 athletes (35%).
The posterior recess angle varied between 28° and 148° for the controls and between 18° and 160° for the internal impingement/GIRD athletes. The mean posterior recess angle measured 65° (SD 27°) for the controls and 94° (SD 38°) for the athletes (P = 0.002). The angle measured greater than 90° in four and six of the 26 controls (15% and 23%) for readers 1 and 2, respectively. Both readers measured an angle greater than 90° in 14 of the 26 internal impingement/GIRD athletes (54%).
Twelve of the 26 internal impingement/GIRD athletes had a deficit of at least 25° in internal rotation, or six vertebral body levels on the scratch test, relative to the contralateral shoulder. The average age of these 12 athletes was 21 years, versus 24 years for the 14 athletes with milder GIRD. The mean labral length was 5.9 mm for the more severe GIRD subgroup and 6.8 mm for those with a milder deficit on examination; the ‘thick-capsule labrum length’ was 8.1 mm vs 9.3 mm, and the posterior recess angle was 93° vs 95°, respectively (all P > 0.5).
Nine of the 26 patients went to surgery. At arthroscopy, four patients had a posterior type 2 SLAP tear, three had posterosuperior labral fraying, and two had a normal labrum. Four patients had a partial thickness tear of the posterior supraspinatus or infraspinatus tendon, one had “synovitis over the posterior cuff”, and, in four, the cuff was normal. All nine patients had either labral or rotator cuff pathology posterosuperiorly. Posterior capsular release and posterosuperior cuff and/or labral debridement, all of which occurred later in the study period, were performed in three patients. Surgical procedures in the other six were posterosuperior debridement of the cuff or labrum only in four, anterior capsulorrhaphy in one, and posterior labral tear repair in one. The remainder improved with physical therapy that included posterior inferior capsular stretching.
Four of the control group patients went to surgery. Two underwent a Mumford procedure for acromioclavicular joint pain, and two had an anterior capsulorrhaphy. Of the other 22 patients, 12 were asymptomatic at the most recent clinic visit, five were improving with physical therapy, two had their care transferred to a chronic pain clinic, and three were not seen in follow up.
The mean humeral head rotation was externally rotated + 24° (range +5° to +49°, SD 11°) in the control group, and + 23° (range +2 to +54°, SD 14°) in the GIRD group.
The Pearson product moment correlation between the two readers was 0.71 for the labral length (95% confidence interval 0.45, 0.86). For the ‘thick-capsule labrum length’ it was 0.74 (0.48, 0.87), and for the posterior recess angle it was 0.69 (0.39, 0.88). The mean absolute percent difference was 0.049 for the labral length (95% confidence interval 0.036, 0.061). For the ‘thick-capsule labrum length’ it was 0.060 (0.046, 0.073), and for the posterior recess angle it was 0.082 (0.063, 0.107). The AUC for labral length was 0.751 (standard error 0.070) for reader 1 and 0.731 (0.070) for reader 2 (P = 0.73). For ‘thick-capsule labrum length’ the AUC was 0.805 (0.066) for reader 1 and 0.811 (0.061) for reader 2 (P = 0.91). For the posterior recess angle, the AUC was 0.710 (0.075) for reader 1 and 0.689 (0.077) for reader 2 (P = 0.78).
For intra-observer variability, the Pearson product moment correlation between the two sets of measurements by reader 2 was 0.82 for the labral length (95% confidence interval 0.63, 0.92). For the ‘thick-capsule labrum length’ it was 0.86 (0.71, 0.94), and for the posterior recess angle it was 0.87 (0.73, 0.94). The mean absolute percent difference was 0.047 for the labral length (95% confidence interval 0.037, 0.057). For the ‘thick-capsule labrum length’ it was 0.051 (0.039, 0.063), and for the posterior recess angle it was 0.055 (0.041, 0.069). These values were considered by our biostatistician to show reasonable inter-observer and intra-observer reliability.
Painful internal impingement in overhead-throwing athletes was first described by Walch et al.  and later expanded upon by Jobe. [7, 8]. In patients with the clinical diagnosis of pathologic internal impingement, fraying or tearing of the posterosuperior labrum and/or posterior cuff is seen at arthroscopy in 81–100% [9, 12, 13, 19]. These lesions can be difficult to see at MR imaging, although they are more apparent on MR arthrographic images including ABER position images [12, 13].
More recently, Burkhart et al. proposed that the primary insult in patients with pathologic internal impingement is a tight posteroinferior capsule from repetitive microtrauma, causing the posteroinferior capsule and posterior band of the IGL to thicken and contract . They hypothesized that the tight posterior capsule combined with anterior capsular stretching is a more likely explanation for the repetitive abrasion injury seen in the posterior superior labrum and rotator cuff in throwers. These structural changes to the capsule could also create shear stresses to the posterior supraspinatus and infraspinatus tendon, leading to an articular surface partial thickness cuff tear and allowing excessive twisting of the biceps tendon, causing a ‘peel-back’ SLAP tear of the posterior superior labrum .
Although a thickened and contracted posteroinferior capsule has been noted at arthroscopy in some elite long-standing professional baseball pitchers with GIRD, there is little in the literature as to whether anatomic changes at the same location are also present in other overhead-throwing athletes. Internal impingement pain and GIRD are known to occur in younger baseball pitchers, position baseball players, and tennis players [6, 10, 14]. Our study finds that there is a statistically significant difference in several posterior inferior labrocapsular measurements between overhead-throwing athletes with painful internal impingement and GIRD and a control group of non-throwers.
One measurement that we made was the length of the posterior inferior labrum at the 8 o’clock position from the glenoid rim to the labral tip. The posterior band IGL inserts onto the labrum in this region, and the ligament contributes collagen fibers to the circumferential collagen fibers of the labrum . We decided to measure the labrum, rather than just the ligament or capsule, for three reasons. First, the labrum in this region should experience similar forces during the deceleration phase of throwing, because of the continuity of collagen fibers, and it might also be expected to thicken from the repetitive microtrauma. Second, the posterior band IGL was often difficult to separate from the capsule near the attachment onto the labrum, so a thickness measurement of the ligament would be difficult to obtain. An isolated stretch of the posterior band IGL can be depicted with the arm in the ‘sleeper stretch position’, which may make it more visible, but we did not image any of our patients in this position . Third, although we noted that the capsule was often thicker in our overhead-throwing athletes, it was still usually only a few millimeters thick. It also tapered as it extended laterally so we would have had to arbitrarily choose a distance from the glenoid attachment at which to measure the thickness. In reviewing the images of our overhead throwers, we found that none had the type of 6 mm thick capsule described in the symptomatic elite professional baseball pitchers .
We found that the posteroinferior labrum is thicker even in younger overhead athletes and non-baseball pitchers with internal impingement pain and GIRD than in our controls. We are uncertain if this contributes directly to the internal rotation deficit or if it is a secondary finding. We also do not know if it is also present in asymptomatic overhead-throwing athletes and whether it decreases toward normal with appropriate stretching exercises.
Another lesion in overhead throwers that occurs in the same region is the Bennett lesion, which is enthesopathy or heterotopic ossification near the insertion of the posterior band IGL [21, 22]. We are not certain if the labrocapsular thickening we describe is a different response to the same repetitive injury or if some of these patients with GIRD may go on to develop a Bennett lesion if it is untreated. Some of the CT arthrogram images of athletes with a Bennett lesion in prior articles do appear to show a thick posterior labrum and non-distended posterior recess [21, 22].
Some of our controls also had a thickened labrum, measuring up to 9.5 mm. Glenoid dysplasia, or a hypoplastic posterior glenoid, is associated with thickening of the posteroinferior labrum as well as instability and posterior labral tears [23, 24]. We did not specifically assess the shape of the posterior glenoid in our study. Although none of our athletes had instability, and none of the athletes or controls had a posterior labral tear, mild hypoplasia may have been the cause of the thickened labrum in some of our patients.
We also found an increased ‘thick-capsule labrum length’ in our internal impingement/GIRD overhead-throwing athletes. Patients with a greater ‘thick-capsule labrum length’ typically had a combination of a thick labrum, a capsule that was thickened near the labral insertion site, and a contracted labrocapsular recess. Although this combination was more common in the overhead throwers, present in 58–65%, a similar finding was seen in 12–19% of the control patients.
The final measurement that we made was the posterior recess angle. This was an attempt to objectively measure the contracted posteroinferior recess seen at arthroscopy in symptomatic elite professional baseball pitchers. This may be the most important finding in GIRD, because stretching exercises may be able to lengthen a tight posteroinferior capsule and re-establish a recess, while relieving their symptoms, but the capsule might remain thickened on MR images. A tight capsule with a constricted posterior labrocapsular recess should indicate persistent GIRD, however, which might be expected to cause continued internal impingement pain.
We found that only 54% of the internal impingement/GIRD athletes had a significantly contracted posterior recess, as defined as a posterior recess angle greater than 90°, compared with 15–23% of the controls. The mean angle was also not significantly different in the overhead throwers with more severe GIRD as compared to that in the athletes with a milder internal rotation deficit.
There are several possible reasons for why this measurement was not a better discriminator between the two groups and why 46% of the internal impingement/GIRD athletes had a well-formed posterior recess (labrocapsular angle less than 90°). First, we did not rigidly standardize the degree of external rotation of our patients’ shoulders during the arm-at-the-side portion of the MR scan. Although we try to position all our patients in relaxed mild external rotation, and the average humeral head rotation in the two groups was similar, review of the images showed that some of the athletes were in fairly marked external rotation. This may have been the most comfortable ‘neutral’ position in these throwers with a lax anterior capsule and an internal rotation deficit. The marked external rotation may have allowed the posterior capsule to relax enough so that a well-formed recess was present at MR arthrography and the posterior recess angle measured less than 90°. We did not attempt to measure the size or depth of the recess, but this may be a better measurement. Finally, some patients with early GIRD may have had a stiff non-compliant posterior capsule yet one which did not appear significantly contracted on the MR arthrogram images.
Of the control patients, 15–23% did not have a deep well-formed posterior recess. Review of the images showed that some of these control patients were almost in internal rotation, which may have been more comfortable within the tight confines of the MR scanner. This would tighten the posterior capsule and could eliminate the recess and cause an obtuse labrocapsular angle.
There also may have been differences in the amount of fluid localized to the posteroinferior region within the joint. We typically did not aspirate joints until they were dry, so there may have been varying amounts of pre-existing effusion. Although we adjusted the volume injected slightly (between 12 ml and 15 ml) for the size of the patient, this was not done following a strict formula, and we would stop at 12 ml if the distension was causing discomfort.
There are several additional weaknesses of this study. Our study population was selective, because many patients with painful internal impingement and GIRD are not imaged with MR arthrography, especially if they improve with physical therapy. Second, we did not look at overhead-throwing athletes with painful internal impingement who did not have significant GIRD (<10° difference in internal rotation or ≤1 vertebral body on the back scratch test). We do not know if these overhead throwers also have a similar morphology to their posterior inferior labrocapsular complex or if there is a different biomechanical explanation for their internal impingement pain. Third, two internal rotation tests were used, and these were performed by three different examiners, so the selection criteria were not rigidly standardized. Fourth, although the controls were not overhead-throwing athletes when they presented with shoulder pain, we did not determine if any had been involved in athletics when they were younger. Fifth, there was overlap in the measurements, so there is no accurate cutoff value between the healthy and GIRD groups. Finally, biceps tendon disease can cause an internal rotation deficit on physical examination due to pain, particularly on the ‘back scratch test’, and we did not attempt to exclude these patients .
In summary, overhead-throwing athletes with pathologic internal impingement and GIRD tend to have an increased labral length and a shallower posterior capsular recess near the attachment of the posterior band IGL. In many, the posteroinferior capsule is also thickened, giving an increased ‘thick-capsule labrum length’. These findings may contribute to the internal rotation deficit present in these overhead throwers. In the absence of a hypoplastic glenoid, these may be useful secondary signs of pathologic internal impingement and should alert the radiologist to closely inspect the posterior cuff and posterosuperior labrum for tears.